OpenRouter has published a comparison of free LLM APIs across providers including OpenRouter, Groq, Gemini, and others. The guide focuses on practical limits such as rate caps, model availability, real costs, and data-use policies.
That is useful because “free” AI access often works for demos but breaks down when a side project starts seeing real usage. Developers need to understand throughput, reliability, and upgrade paths before building around a provider.
The post reflects a broader developer concern: model access is now an infrastructure choice, not just a pricing checkbox.